The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes user-supplied markdown files to extract tasks and dependencies, which represents an indirect prompt injection surface. Ingestion points: Loads and parses content from 'docs/features/[feature-name]/breakdown.md' as specified in the SKILL.md instructions. Boundary markers: There are no instructions defining the use of delimiters or 'ignore embedded instructions' warnings when the agent reads the breakdown data. Capability inventory: The skill includes a 'workflow-validator.py' script for local file system reads and provides instructions for the agent to create and organize project files. Sanitization: The validator script checks for document structure using regex but does not inspect or sanitize the actual text content for malicious instructions.

feature-planning